Are We Automating the Wrong Parts of Sales?
About this episode
In Episode 4 of Intent, Decoded, Brendan Hughes is joined by Mark Colgan, who builds and hands over AI-enabled outbound systems for B2B SaaS teams at Yellow O, and Niall O'Gorman, founder of MarketSizer. Mark's mantra is to automate the process and master the conversation, and most of the episode is spent working out where that line sits. He calls the AI SDR wave premature automation: scaling a process before anyone has proven it works, so a 0.1% reply rate on 500 emails becomes a 0.1% reply rate on 5,000. His starting point is closed-won and closed-lost analysis, then interviews with the reps, because what the process says and what reps actually do are rarely the same thing. Niall brings the data side: dirty CRM records, transcription tools writing the wrong next steps, and AI that amplifies whatever foundations it is pointed at. The conversation covers human in the loop, fake personalisation, company and playbook context for AI agents, feedback loops from prospects who say no, what stays human in selling, AI call analysis for coaching, why AI is closer to ready for inbound than outbound, the bottlenecks that appear when outbound starts working, cold email as a nurture channel, and knowing when not to act on a signal.
Topics covered
- Premature automation: scaling a process before it is proven
- Why ten times the volume at a 0.1% reply rate is still 0.1%
- Starting with closed-won and closed-lost analysis
- The gap between the documented process and what reps actually do
- Dirty CRM data, and why AI amplifies it
- Human in the loop: where to speed up and where to slow down
- Fake personalisation and the wrong football club
- Company-level and playbook-level context for AI agents
- Feedback loops from the prospects who say no, or say nothing
- What stays human: relationships, emotion and judgement
- AI call analysis for coaching and objection handling
- Why AI is closer to ready for inbound than for outbound
- Grouping dialler call blocks by persona and angle
- When outbound success creates a bottleneck downstream
- Cold email as a nurture and distribution channel
- Knowing when not to act on a signal
Chapters
- 00:00Introduction: Are We Automating the Wrong Parts of Sales?
- 05:24Why Premature Sales Automation Fails
- 08:20Closed-Won vs Closed-Lost Sales Analysis
- 13:48The Hidden Friction in Sales Automation
- 16:13Human-in-the-Loop AI and Sales Prospecting
- 21:58Context Engineering for AI Sales Agents
- 26:49What Should Stay Human in Sales?
- 31:32Using AI for Sales Coaching and Objection Handling
- 35:37Inbound vs Outbound AI Sales Automation
- 40:16When AI-Powered Outbound Creates Bottlenecks
- 44:25Cold Email, Nurturing and Buyer Relationships
- 47:40Buyer Intent, Competitive Signals and Outreach Timing
Key takeaways
- Prove the process before you automate it. A 0.1% reply rate on 500 emails will be a 0.1% reply rate on 5,000. Automation multiplies whatever is already there, including the targeting, the message and the evidence.
- Start with the deals you lost, not only the ones you won. Closed-lost analysis shows whether you lost on product, on timing or on something else, and a failed deal often teaches more than a successful one.
- Interview the reps. What the process says, what RevOps built and what reps actually do every day are three different things, and the gap usually includes a shaky grasp of the persona.
- AI amplifies the foundations it is pointed at. Duplicate CRM records, empty fields and transcription tools writing the wrong next steps compound with every system and person downstream.
- Speed up the research and slow down the conversation. Give the rep an AI research brief with five or six angles, then let them choose one. Ticking a personalisation box is not the same as being relevant.
- Write the context down, even if it is ugly. A matrix of segments, personas, pains, value and proof is for the AI agents to read, not for you. Add playbook context too, so a rep knows why a list exists before working it.
- AI is closer to ready for inbound than outbound. When the buyer reaches out, they mostly want an answer. When you reach out, tone, timing and the person on the other end matter more than any agent can yet handle.
- Knowing when not to reach out is a discipline. Signals are plentiful. Hitting a perfectly good opportunity too often with the wrong message burns it, so someone in the organisation has to own the restraint.
Moments worth quoting
“My mantra is automate the process and master the conversation.”
“I call this premature automation. Essentially, we're trying to automate something before we know it's a proven process.”
“If you've got dirty data within your CRM and you point a very sophisticated AI at dirty data, it's just going to make a lot of noise.”
“It doesn't matter that it's ugly. It doesn't matter that it looks overwhelming, because you're not the one reading it.”
“You've knocked on a hundred doors, 99 have said no, and you're still able to put a smile on your face and figure out a way to cut through the noise. I think that's the nuance which is still missing from an AI solution.”
“Knowing when not to reach out is a practice in and of itself.”
Mentioned in this episode
- Read the full recap - The written companion to this episode, with the takeaways and the sharpest quotes.
- Why CRM Data Is Inaccurate - The dirty-data problem Niall describes, and why AI makes it louder rather than quieter.
- Sales Outreach Not Getting Responses? - What to fix before you add volume, the premature automation problem from the other side.
- B2B Winback: Recover Churned Customers on Competitor Renewal Timing - The timing play behind Mark's closing tip on competitor tenure and churn.
- From Signal to Trust: How Sellers Earn the Right to Reach Out - The Episode 2 recap with Leslie Venetz, on relevance over fake personalisation.
- When Every Tool Has the Same Signals - The Episode 3 recap, on why signals multiply an outbound motion rather than rescue one.
- Yellow O - Mark's practice, building AI-enabled outbound systems for B2B SaaS teams.
- Start your free trial - See which accounts are showing observable signs of movement, from competitor trials to renewal windows.
Keep listening
Intent Data Is Dead. Or Is It? Two Signal Companies Argue It Out
Two signal companies, one common enemy. Jack Porter of WhiteWhale joins Niall O'Gorman and Brendan Hughes to work out what still counts as a buying signal once every tool ships the same ones.
From Signal to Trust: How Sellers Earn the Right to Reach Out
A signal is not permission to sell. Leslie Venetz joins Niall O'Gorman and Brendan Hughes to unpack what should actually happen after a GTM team sees an intent signal - and how to earn the right to a buyer's time.
Why most GTM teams are misusing intent data
Buying intent data is everywhere, yet most teams still struggle to turn it into pipeline. We unpack the practical mistakes GTM teams make when reading intent signals, why category-level spikes rarely translate to closed revenue, and how to combine intent with account fit and timing to actually book meetings.
Find out who's evaluating in your market right now
before your competitors do.
30 minutes. No commitment. We'll show you the accounts actively evaluating in your market today.
No credit card · Tailored to your ICP · Live data, not a slide deck